Spectrum Scale is a highly scalable, high-performance storage solution for file and object storage. IBM continues to enhance Spectrum Scale performance, in response to recent hardware advancements and evolving workloads. This presentation will discuss performance related improvements in Spectrum Scale 5.X versions focusing on enhancements made in support of AI and HPC use cases, including improvements for IO 500 performance benchmark, improvements to RDMA to select NUMA aware adapters and more.
Download slides here
Recordings will be made available soon
User Group Host: Kristy Kallback-Rose
Speakers:
| Speaker Name | Photo | Bio |
| John Lewars |  | John Lewars is a Senior Technical Staff Member leading performance engineering work in the IBM Spectrum Scale development team. He has been with IBM for over 20 years, working first on several aspects of IBM's largest high performance computing systems, and later on the IBM Spectrum Scale (formerly GPFS) development team. John's work on the Spectrum Scale team includes working with large customer deployments and improving network resiliency, along with co-leading development of the team's first public cloud and container support deliverables. |
| Jay Vaddi |  | Jay Vaddi is a Storage Performance Engineer at IBM Tucson, AZ. He has been with IBM and the performance team for over five years. His focus is primarily on performance analysis and evaluations of IBM Spectrum Scale and IBM Elastic Storage System products. |
| Olaf Weiser |  | Olaf works with GPFS for over 15 years now. He started his GPFS career in one of the worlds biggest telecommunication companies as a technical administrator. Since more than 10 years, Olaf is with IBM as storage consultant and performance specialist. Recently, he joined IBM Research and Development and works on enhancements in Spectrum Scale to adopt client and customer's needs in the product. |
| Pidad D'Souza |  | Pidad D'Souza is a System Architect specializing in performance engineering of IBM Spectrum Scale and IBM POWER systems. He has been with IBM for over 17 years, leading system performance engineering of GPU accelerated systems designed for applications in the fields of AI & HPC. He has presented extensively at several international conferences, and delivered customer workshops and lab sessions. |
With GPUDirect Storage (GDS) NVIDIA has created a data path from the storage directly into the buffer of NVIDIA GPUs. IBM Spectrum Scale is perfectly suited to take advantage of this feature from NVIDIA to further enhance storage performance. This joint session by NVIDIA and IBM development provides GDS overview, highlights how GDS has been integrated with Spectrum Scale and how it can be used to drive higher performance.
Download slides here
User Group Host: Kristy Kallback-Rose
Speakers:
| Speaker Name | Photo | Bio |
| Ingo Meents |  | Ingo Meents has been working as an architect in IBM Spectrum Scale and its predecessors’ development for more than 10 years. His current focus is on RDMA technologies like Nvidia’s GPU Direct Storage (GDS) to eliminate the IO bottleneck faced by modern AI applications. He is leading the agile team that has made GDS in Spectrum Scale generally available and is working on further enhancements. |
| Kiran Kumar Modukuri |  | Kiran Modukuri is a Principal software engineer at NVIDIA DGX Platform software team with special focus on accelerating IO pipelines for AI and Machine Learning. He is the Co-architect of NVIDIA MagnumIO GPUDirect Storage and he has technical contributions to Distributed Filesystems, caching and replication technologies over the last 16 years. Kiran has Master's degree in electrical and computer engineering from university of Arizona. |
Starting with IBM Spectrum Scale 5.1.1, Multiple Connections Over TCP (MCOT) is enabled by default which enhances the IBM Spectrum Scale communication subsystem to establish multiple TCP connections for communication between each pair of Spectrum Scale daemons (mmfsd). Using more connections may improve performance and resiliency, particularly for some bonded configurations which can use more than one physical interface per node with multiple TCP connections. Depending on factors like network speed and number of nodes, the value of the new maxTcpConnsPerNodeConn parameter may need to be adjusted up or down.
Download slides here
Recordings will be made available soon
User Group Host: Paul Tomlinson
| Speaker Name | Photo | Bio |
| Jay Vaddi | | Jay Vaddi is a Storage Performance Engineer at IBM Tucson, AZ. He has been with IBM and the performance team for over five years. His focus is primarily on performance analysis and evaluations of IBM Spectrum Scale and IBM Elastic Storage System products. |
IBM continues to enhance Spectrum Scale functionality and performance,
in response to recent hardware advancements and evolving workloads.
This talk will discuss key enhancements delivered with Spectrum Scale
5.1.2 including upcoming high performance S3 interface, Spectrum Scale
ECE enhancements, AFM enhancements, performance enhancements and
latest updates on Spectrum Scale BDA and CNSA/CSI.
Download slides here
User Group Host: Paul Tomlinson
Speakers:
| Speaker Name | Photo | Bio | Social connections |
| Chris Maestas |  | Chris is an Executive Architect for IBM File and Object Storage Solutions with over 25 years of experience deploying and designing IT systems for clients in various spaces. He has experience scaling performance and availability with a variety of file systems technologies. He has developed benchmark frameworks to test out systems for reliability and validate research performance data. He also has led global enablement sessions online and face to face where discussing how best to position mature technologies like Spectrum Scale with emerging technologies in Cloud, Object, Container or AI spaces. | Twitter: @cdmaestas
LinkedIn: https://www.linkedin.com/in/cdmaestas |
Restarting for 2022, the SSUG::Digital talks are back with talks over two days.
Day 2 Talk 1: GPU Direct Storage with Spectrum Scale
With GPUDirect Storage (GDS) Nvidia has created a data path from the storage directly into the buffer of Nvidia GPUs. IBM Spectrum Scale is perfectly suited to take advantage of this feature from Nvidia to further enhance storage performance. This joint session by Nvidia and IBM development provides GDS overview, highlights how GDS has been integrated with Spectrum Scale and how it can be used to drive higher performance.
Day 2 Talk 2: Spectrum Scale performance improvements
Spectrum Scale is a highly scalable, high-performance storage solution for file and object storage. IBM continues to enhance Spectrum Scale performance, in response to recent hardware advancements and evolving workloads. This presentation will discuss performance related improvements in Spectrum Scale 5.X versions focusing on enhancements made in support of AI and HPC use cases, including improvements for IO 500 performance benchmark, improvements to RDMA to select NUMA aware adapters and more.
See >>here<< for sessions of Day 1 – Jan 18th, 2022.
User Group Host: Kristy Kallback-Rose
Speakers:
| Speaker Name | Photo | Bio |
| Ingo Meents | | Ingo Meents has been working as an architect in IBM Spectrum Scale and its predecessors’ development for more than 10 years. His current focus is on RDMA technologies like Nvidia’s GPU Direct Storage (GDS) to eliminate the IO bottleneck faced by modern AI applications. He is leading the agile team that has made GDS in Spectrum Scale generally available and is working on further enhancements. |
| John Lewars (IBM) | | John Lewars is a Senior Technical Staff Member leading performance engineering work in the IBM Spectrum Scale development team. He has been with IBM for over 20 years, working first on several aspects of IBM's largest high performance computing systems, and later on the IBM Spectrum Scale (formerly GPFS) development team. John's work on the Spectrum Scale team includes working with large customer deployments and improving network resiliency, along with co-leading development of the team's first public cloud and container support deliverables. |
| Jay Vaddi | | Jay Vaddi is a Storage Performance Engineer at IBM Tucson, AZ. He has been with IBM and the performance team for over five years. His focus is primarily on performance analysis and evaluations of IBM Spectrum Scale and IBM Elastic Storage System products. |
| Kiran Kumar Modukuri | | Kiran Modukuri is a Principal software engineer at NVIDIA DGX Platform software team with special focus on accelerating IO pipelines for AI and Machine Learning. He is the Co-architect of NVIDIA MagnumIO GPUDirect Storage and he has technical contributions to Distributed Filesystems, caching and replication technologies over the last 16 years. Kiran has Master's degree in electrical and computer engineering from university of Arizona. |
| Pidad D'Souza | | Pidad D'Souza is a System Architect specializing in performance engineering of IBM Spectrum Scale and IBM POWER systems. He has been with IBM for over 17 years, leading system performance engineering of GPU accelerated systems designed for applications in the fields of AI & HPC. He has presented extensively at several international conferences, and delivered customer workshops and lab sessions. |
